Research and resources
Get new Metano Labs research when it publishes: real attacks on AI agents, the mechanism behind each, and what stops them.
- Research· 4 min read
Beyond Maliciousness: Attack Surface Profiling of Agent Skills
An agent skill runs with everything the agent that loads it can do. New research from Metano MARS Labs, Purdue University and VulNow profiles more than 6,000 skills sourced from a couple of different marketplaces, to measure how much of that authority they put into play.
Japneet Singh
- Integrations· 3 min read
Metano Integrates with Okta
Metano is live on the Okta Integration Network. Tie every AI agent to its Okta identity, see what it does at runtime, and stop the agent and its access in one step.
Varun Anand
- Research· 15 min read
Poetry as a Dead Drop
Operation Poetic Injustice: your LiteLLM gateway is somebody else’s command-and-control server.
Ratnesh Pandey
- Research· 8 min read
Claimed, Not Enforced: The MCP Security Gap
From broken authorization to skipped confirmations, Metano’s MCP research found one recurring problem: the control isn’t where the action happens.
Parth Shukla
- Product· 3 min read
Kill Switch
Autonomous agents need autonomous-era containment. Meet the AI kill switch.
Varun Anand
- Product· 4 min read
Shadow AI Visibility
Shadow AI starts with visibility, but securing it means understanding what agents actually do.
Shravanthi Reddy
- Research· 8 min read
Sleeper Skills
A marketplace audit score attests to the artifact that was scanned, not to what a skill becomes after a silent auto-update. Across several skill marketplaces, 50+ skills can update themselves. Together they have been downloaded more than 28,000 times and starred more than 23,000 times.
Harsh Akshit
- Research· 11 min read
The Ghost in the Skills
Public AI agent skills instruct agents to install packages that were never published. The names sit unregistered, allocation is first come first served, and the agent does not hesitate. A supply chain attack nobody had to break in to stage.
MARS Labs (Metano Ai Research & Security Labs)
- Research· 7 min read
The AI-Powered Persistent Threat (AIPT)
How AI is turning the Advanced Persistent Threat into the AI-powered Persistent Threat, and what defenders have to change.
Japneet Singh
- Research· 6 min read
Why Your SIEM Has No Idea What Your AI Agents Did
Your security stack logs almost everything. The one thing it cannot reconstruct is the decision an agent made, and why it made it.
Shravanthi Reddy
- Product· 8 min read
Skills Lie. We Stop That.
A skill’s description is a promise to your agent. Its behavior is the part nobody checks. Every other scanner reads what a skill says it does. SkillTracer runs it and reports what it actually did.
Japneet Singh
- Research· 6 min read
Your EDR Can’t See Agent Attacks
Agents are being weaponized and hijacked, and the orchestration that makes them dangerous is invisible to the telemetry your SOC runs on.
MARS Labs (Metano Ai Research & Security Labs)
Showing 1 - 12 of 12 posts